Atomics Reduce Counting Optimization for SplitK Skinny GEMMs. (#29843)

Signed-off-by: Hashem Hashemi <hashem.hashemi@amd.com>
This commit is contained in:
Hashem Hashemi
2026-01-16 11:45:04 -06:00
committed by GitHub
parent 9fd918e510
commit 7a1030431a
6 changed files with 635 additions and 10 deletions
+6
View File
@@ -26,6 +26,12 @@ TORCH_LIBRARY_EXPAND(TORCH_EXTENSION_NAME, rocm_ops) {
"Tensor");
rocm_ops.impl("wvSplitK", torch::kCUDA, &wvSplitK);
// Custom gemm op for skinny matrix-matrix multiplication
rocm_ops.def(
"wvSplitKrc(Tensor in_a, Tensor in_b, Tensor? in_bias, int CuCount) -> "
"Tensor");
rocm_ops.impl("wvSplitKrc", torch::kCUDA, &wvSplitKrc);
// wvSplitK for fp8
rocm_ops.def(
"wvSplitKQ(Tensor in_a, Tensor in_b, Tensor? in_bias, Tensor! out_c, "